fin.
If you made it this far, thanks for reading!
There's loads more in the paper if you like what you see.
I'm really keen to hear what people think so don't be shy and leave a reply/shoot me a dm.
www.biorxiv.org/content/10.6...
6/n
I thought this result was quite surprising since it cuts against the grain of popular predictive processing narratives and suggests task priors are doing more for the maintenance and transformation of sensory representations into action than for sensory encoding per se.
Neural tuning to motion direction expressed as decoding accuracy stratified by A) stimulus precision and B) prior precision during the observation and response periods (note the difference y-axis scales). Vertical red lines and grey dashed lines indicate coherent motion onset, and offset, respectively. Black solid, dashed and dotted lines indicate cluster significance p < .05 of pairwise condition comparisons. For the purposes of presentation, each trace was smoothed using a temporal Gaussian filter (sigma = 4). All statistical comparisons were performed on the unsmoothed data.
5/n
Inverted encoding models showed strong effects of motion coherence on representational fidelity during stimulus observation which carried over into the response period (top).
Priors also enhanced stimulus representation (bottom), but only in the response period once the stimulus was gone! 🤯
4/n
How does this bias emerge?
Recent work suggests task priors might accelerate evidence accumulation by directly enhancing early neural representations of more likely stimuli.
We were able to characterize these neural representations using EEG recorded during the task. 🧠
Results next!
Behavioural results for Experiment 2 (EEG + Behaviour). A) Trial-level error histograms for each level of stimulus precision. B) Response accuracy stratified by stimulus precision and prior precision. C) Response bias stratified by stimulus precision and prior precision. Bias expressed as mean signed-error magnitudes such that attraction to the central-tendency of the current prior is captured by a positive shift in the error distribution. D) Bias stratified by stimulus precision and time within the session grouped into three equal-sized bins. Here, time reflects the exposure participants had to each of the priors. For visualisation, the data used to construct boxplots were normalised to remove the between subjects’ variability (Cousineau, 2005). Inset bar plots indicate t-statistics and significance of mixed GLM parameter estimates. * = p < .05.
3/n
Estimation behaviour followed the textbook Bayesian pattern:
Responses were biased towards expected motion directions and the magnitude of bias scaled with the relative precision of the stimulus and the prior. (panel C)
Weak stimulus + strong informative prior = biggest bias.
Overview of the task and the manipulation of stimulus and prior precision. A) Structure of a single trial from pre-fixation to response. Note: trial diagram is schematic only, colours and scale differ from actual design (see Methods). B) Stimulus precision was manipulated via motion coherence, corresponding to the precision of the sampling distribution of possible dot-directions on each frame. Zero corresponds to the global target motion direction. C) Priors for motion direction were manipulated via the block-level sampling distribution for global target motion directions. These distributions could either be completely uninformative (uniform) or contain some information about the probability of upcoming stimulus values. In this case zero corresponds to the central tendency of the block-wise sampling distribution/prior.
2/n
Participants made continuous decisions about the direction of a dot-motion stimulus with varying levels of coherence. We manipulated prior expectation by sampling the target directions from von Mises distributions whose location and concentration parameters varied each block.
🏰 New preprint! 🏰
(w/ the excellent @neuronomist.bsky.social & Prof. Jason Mattingley)
For my PhD I've been using EEG, psychophysics and modelling to explore how the brain uses uncertainty to balance sensory information against expectations about what's out there.
🧵1/n
Late Integration of Prior Expectations During Precision Weighted Perceptual Decisions www.biorxiv.org/content/10.64898/2026.04...
Oh to be a little Tardigrade walking across a microscope slide. 🫧🐻🧪
Slide from Tim Cottier's presentation which says 'Super recognisers are more sensitive to expression' with plots showing the results
Fantastic presentation from @tvcottier.bsky.social kicking off our first meeting for 2026. His recent work suggests super-recognisers prioritise identity cues in faces over other dimensions, & that expressive cues interfere with them - suggesting these processes are not independent
Second up was @tgastrell.bsky.social whose research explored whether neural tuning depends on the precision of learned priors using a visual dot motion paradigm...
Excited to be sharing some recent work of mine with this fantastic group!
Photos of the seven-strong committee for the Brisbane Experimental Psychology Student Initiative
Introducing @brisepsi.bsky.social - we're an experimental psychology initiative run by grad students in Brisbane (Meanjin) across UQ & QUT 🧠
We host a monthly dose of freshly baked research (students, ECRs, & bigwigs) followed by social goodness. Help us grow with a follow! 🤗 #neuroskyence
Preventing data leakage in neural decoding - "for autocorrelated neural time series, standard k-fold cross-validation can dramatically overstate performance." www.biorxiv.org/content/10.6...
And it's out now in Cortex: www.sciencedirect.com/science/arti...
Summary below 🧵
Excited to be presenting new EEG results from my PhD at this years @acnsau.bsky.social annual meeting in Melbourne.
If you're interested in how priors affect multivariate stimulus representations in the brain come through to the Sensation and Perception stream this afternoon!
#ACNS2025
New paper in Imaging Neuroscience by Margaret Jane Moore, Amanda K. Robinson, and Jason B. Mattingley:
Expectation dynamically modulates the representational time course of objects and locations
doi.org/10.1162/IMAG...
The final bit of work from my PhD just got published at JOV! We looked at similarity judgements made for naturalistic image patches, and whether these are predicted by simple image statistics… (spoiler: yep!)
Link to paper: doi.org/10.1167/jov....
1/11
Just published some work at Scientific Reports! We investigated visual adaptation following free viewing of a film (Casablanca) that had its oriented contrast altered. To our surprise, we found adaptation effects to be pretty negligible…
www.nature.com/articles/s41...
1/10
🚨Pre-print of some cool data from my PhD days!
doi.org/10.1101/2025...
☝️Did you know that visual surprise is (probably) a domain-general signal and/or operates at the object-level?
✌️Did you also know that the timing of this response depends on the specific attribute that violates an expectation?
My final Registered Report from my PhD is now out in Cortex!
We investigated the role of dopamine & HD-tDCS in mind wandering & sustained attention. While stimulation had no effect, increasing dopamine reduced spontaneous thought and may protect against performance deficits. doi.org/10.1016/j.co...
I really like this paper. I fear that people think the authors are claiming that the brain isn’t predictive though, which this study cannot (and does not) address. As the title says, the data purely show that evoked responses are not necessarily prediction errors, which makes sense!
Well said. Love this take.
Excited to be presenting behavioral and modeling work looking at perceptual decision making under uncertainty with recently learned priors at the EPC/APCV joint meeting today at UNSW, Sydney!
Catch me at 2:45pm in Gallery 1!
@expsyanz.bsky.social
#epc2025
5/5 So the mystery of mechanism remains, but the take home message is this:
Visual awareness is not just about the integration of stimulus history with stimulus content - visual field location matters as well!
OA paper (Gastrell et al., 2025, Journal of Vision): 🔗 doi.org/10.1167/jov....
4/5 Next, we asked whether differences in fixational stability between foveal and peripheral viewing might drive the effect?
In a 4th Exp., we replicated the original effect *again* (so this really IS a thing!), but spatial differences in fixational stability did not correlate either.
3/5 We now knew this was a low-level (spatially specific) effect, so in Exp. 3 we replicated it, and also probed motion adaptation - a likely mechanism.
We found that adaptation to our unambiguous prime generated stronger motion after effects in the periphery, but didn't predict SFM effect! 🤯
2/5 In Exp. 2 we abolished the effect by moving the stimulus between v.f. locations as it changed from unambiguous to ambiguous motion.
This ruled out a high-level (non-retinotopic) visual explanation whereby the influence of a prime on object-level representations might depend on its precision.
1/5 Using unambiguous → ambiguous SFM sequences, we tested whether priming effects on perception of bistable structure from motion vary between fixation and the periphery.
In Exp. 1 we show that immediate perception of a target is more biased towards primes when sequences are fixated v peripheral.
New paper @ JOV!
With Matt Oxner, Frank Schumann and David Carmel.
🧠 Perception of ambiguous motion is shaped by recent visual experience - but does this differ across the visual field?
Yes!
Full open access paper here (Gastrell et al., 2025, Journal of Vision): 🔗 doi.org/10.1167/jov....
🧵
